[MagicAndre1981]

iu1nguoi, on 25 March 2012 - 12:40 AM, said:

it is not the problem of Windows but Corsair SSD itself

ok. How did you found the cause?

[Yogurt]

MagicAndre1981, on 24 March 2012 - 02:47 PM, said:

Yogurt, on 24 March 2012 - 12:48 PM, said:

What is the point to check the log if it's clearly not showing the truth?...

which log?

I meant the .etl file, viewed by the Windows Performance Analyzer GUI.

MagicAndre1981, on 24 March 2012 - 02:47 PM, said:

have you created the XML? What does it show?

Of course I've created XML. It starts with
<scenario start="2175477" duration="135916882" suspend="15078000" resumecritical="554043" hiberwrite="69250000" hiberpageswritten="336129" hiberread="29422000" resume="1336000">
And the branches with significant duration are
<suspendapps start="2175643" duration="2280457" totalChildrenDuration="1065107">
<flushvolumes start="4635516" duration="7942940" />
<querydevices start="4491802" duration="143714">
<suspenddevices start="12578456" duration="4675687">
...

I don't know in which units are these (clearly not milliseconds) but if they were microseconds, then the total duration were around 136 seconds (may be true).

And what is interesting, me too have a
<device start="16473356" duration="34737" name="IDE\DiskST9500423AS_____________________________0001DEM1\4&amp;160fd31b&amp;0&amp;0.0.0">
<device start="16127668" duration="344569" name="IDE\DiskSATA_SSD________________________________S5FAM011\4&amp;160fd31b&amp;0&amp;0.2.0">
lines. I'm 100% sure I've set my disks to AHCI mode in BIOS.
msinfo32 shows that Windows uses iastor.sys 10.1.0.1008, that's Intel® Mobile Express Chipset SATA AHCI Controller.

[iu1nguoi]

MagicAndre1981, on 25 March 2012 - 08:14 AM, said:

iu1nguoi, on 25 March 2012 - 12:40 AM, said:

it is not the problem of Windows but Corsair SSD itself

ok. How did you found the cause?

i made a search (which I dont know why I could not do it before) and found a Corsair forum where many customers complaint about this issue. Corsair has also confirmed it, but so far their solution still is unsure yet. So I guess I just have to use hibernate instead.

[MagicAndre1981]

Yogurt, on 26 March 2012 - 01:20 AM, said:

Of course I've created XML. It starts with
<scenario start="2175477" duration="135916882" suspend="15078000" resumecritical="554043" hiberwrite="69250000" hiberpageswritten="336129" hiberread="29422000" resume="1336000">

I don't know in which units are these (clearly not milliseconds) but if they were microseconds, then the total duration were around 136 seconds (may be true).

those are µs. So writing the hibernation file takes 69s and reading takes 29s. This is terrible slow

Yogurt, on 26 March 2012 - 01:20 AM, said:

And what is interesting, me too have a
<device start="16473356" duration="34737" name="IDE\DiskST9500423AS_____________________________0001DEM1\4&amp;160fd31b&amp;0&amp;0.0.0">
<device start="16127668" duration="344569" name="IDE\DiskSATA_SSD________________________________S5FAM011\4&amp;160fd31b&amp;0&amp;0.2.0">
lines. I'm 100% sure I've set my disks to AHCI mode in BIOS.
msinfo32 shows that Windows uses iastor.sys 10.1.0.1008, that's Intel® Mobile Express Chipset SATA AHCI Controller.

this is one of the issues I don't understand. My Samsumg F3 HDD which I run in AHCI shows up as SCSI:

SCSI\DISK&VEN_SAMSUNG&PROD_HD103SJ\

iu1nguoi, on 26 March 2012 - 02:49 AM, said:

i made a search (which I dont know why I could not do it before) and found a Corsair forum where many customers complaint about this issue. Corsair has also confirmed it, but so far their solution still is unsure yet.

this is disappointing for you. That's why I still have a normal HDD. Runs fine and has no issue.

[iu1nguoi]

MagicAndre1981, on 26 March 2012 - 06:53 AM, said:

this is disappointing for you. That's why I still have a normal HDD. Runs fine and has no issue.

that is sad, but true. thank you very much for your help. i do appreciate it

[MagicAndre1981]

You're welcome

[Yogurt]

MagicAndre1981, on 26 March 2012 - 06:53 AM, said:

those are µs. So writing the hibernation file takes 69s and reading takes 29s. This is terrible slow

As I wrote, the hard disk activity is low during hibernation, and it takes this long only if I remove the DC plug during hibernation (otherwise it's less than a minute). But I've benchmarked my drive for a sequential write of 35-45 MB/s, read of 250 MB/s. This would mean a total write of 2.5 GB, and a total read of 7.5 GB.

This snippet is from a plugged-in hibernation:
<scenario start="428100" duration="96392882" suspend="28665000" resumecritical="563837" hiberwrite="23194000" hiberpageswritten="220999" hiberread="20029000" resume="2790000">
Write takes only (?) 23 seconds, read 20 seconds.

MagicAndre1981, on 26 March 2012 - 06:53 AM, said:

this is one of the issues I don't understand. My Samsumg F3 HDD which I run in AHCI shows up as SCSI:

Intel Rapid Storage Technology Manager also swears that my disks are in SATA 3 Gb/s mode with NCQ enabled. (However, it tells my secondary disk is the system disk, and my system disk is not the system disk.)

[TEagle]

Thank You MagicAndre1981 for the brilliant post and instructions. Would you take a look and see what your take is on my slow (7 minutes) win7 shutdown?

http://dl.dropbox.co...ry_shutdown.xml

I think I spotted the culprit at the Logoff section after EndShell with the subscriber named "Sens".

I do not know where to go from there though.. Do you need the ETL file as well?

[MagicAndre1981]

SENS = System Event Notification Service (SENS):

Quote

The System Event Notification Service (SENS) now provides these capabilities in the operating system, creating a uniform connectivity and notification interface for applications. Using SENS developers can determine connection bandwidth and latency information from within their application and optimize the application's operation based on those conditions.

So 1 app waits for events.

I see that Teamviewer hangs:

- <unresponsiveServices numUnresponsiveServices="1">
  <unresponsiveService name="TeamViewer7" /> 
  </unresponsiveServices>

so stop Teamviewer before shutdown. Does this help?

[doveman]

Hi MagicAndre1981

After helping me with my latency problems, perhaps you can help me with my hibernate problems now

I couldn't even get sleep/standby (S3) to work until I disabled Hibernate (powercfg -h off) and then it works fine and if I re-enable Hibernate then Sleep stops working again and the display just blanks but it doesn't standby the PC at all. If I try and Hibernate, not much happens at all. Strangely I'm having the same problem on both a Win7 x64 PC and a Win7 x86 PC.

I've uploaded the .etl here: http://www.mediafire...yvhntooz4ya477t

The summary_hibernation.xml just contains:

<suspend time_unit="us" time_precision="us" min_reported="0">
<scenario start="984746" duration="-0" suspend="-0" resumecritical="-0" resume="-0" />
</suspend>

I wonder if for some reason the System is unable to write to the hiberfil.sys. That file does exist, I'm just wondering if whatever process dumps the RAM contents to it when Hibernating (or Sleeping with Hybrid Sleep enabled) is having some permissions problem and finding itself unable to do so. It's not possible to check the Owner or Permissions on that file as it just says "The requested security information is either unavailable or can't be displayed" but it says the same for pagefile.sys. I'm not sure if that's working either, as the timestamps show it hasn't changed since 23 March, but I think writing to hiberfil.sys in the root of C: probably requires greater authority than writing to pagefile.sys in the the root of E:. I'm not sure how to check whether pagefile.sys is being written to/currently used or when it was last written to/last used other than the timestamps.

EDIT: Ah, seems it could be related to grub4dos and the Windows partition not being Active http://ubuntuforums....d.php?t=1341694 If the Windows partition has to be Active for Hibernate to work, it seems to make it impossible to have Hibernate work with a dual-boot Win7 system though. I've set my C: (which is the 2nd partition on the HDD, the first is my other Win7 install) to Active now and it shows this in various apps, but not Windows Disk Management so I'll probably have to reboot to update this. Hibernate isn't working yet anyway, so hopefully a reboot will fix it.

This post has been edited by doveman: 29 March 2012 - 05:13 AM

[MagicAndre1981]

ok, was grub the cause?

[doveman]

MagicAndre1981, on 29 March 2012 - 06:59 AM, said:

ok, was grub the cause?

Yeah, I've just rebooted and tested Hibernate and it works fine now I've made the current C: Active. I think it will be the same fix on my other PC as that's using grub4dos as well. It'll be nice to finally have Sleep and Hibernate working properly, as I've been trying to get them sorted for a few weeks now.

It's not so much that grub is the cause though, just that my current C: drive wasn't marked Active (this attribute was assigned to the 1st partition on the HD, which is my other Win7 partition). I'll have to check but I don't think there's a way for grub to set the Active flag on the partition the user selects to boot, so it's a case of choosing which of the two Win7 installs you want to have Hibernate working for and marking that partition Active.

[MagicAndre1981]

doveman, on 29 March 2012 - 07:52 AM, said:

Yeah, I've just rebooted and tested Hibernate and it works fine now I've made the current C: Active.

[levelblade]

MagicAndre1981,

Thanks for creating this tutorial. I have applied the approach that you suggest, but I am now stuck and need help.

I have been struggling with very slow resume from hybrid sleep on one of my systems (I have another similar system which resumes very quickly; the difference in resume times prompted me to try to find out why the one system takes so long, and that research led me to this thread.) OS is Windows 7 x64 Professional with 6GB RAM. Per your instructions, I have run a hibernation trace using xbootmgr and similar to some other users, it shows acceptable hiberwrite time (approx. 21 secs) but very, very long hiberread time (more than 276 secs or over 4 minutes). This long delay takes place before devices start to resume as shown by the xml created by xperf (if I am reading it correctly), and no device seems to take an extraordinarily long time to start. (The xml file is attached.)

I have defragged my hiberfil.sys file (by disabling hibernation, restarting, defragging, and then re-enabling hibernation) and have defragged my MFT using third-party software to run a boot-time defragmentation. I have also tried disabling windows search indexer (which apparently helped another user with a similar problem) but that provided no improvement in resume time.

From reading prior posts in this thread, I am guessing that the problem may lie with my hard drive. It is a Hitachi HDS721075KLA330 and its specs (7200rpm and 32mb cache) suggest that it should have plenty of speed such that the hiberread time should not be so slow. I have tried to use ProcMon to confirm that the HDD is at fault, but I am unable to get proper results. Perhaps I am not using ProcMon correctly, but after enabling event capture, I cannot successfully run xbootmgr (from an elevated prompt) if ProcMon is open; I get a warning that the trace cannot be started with an error code of 0x000000b7. I can successfully run xbootmgr only if I close ProcMon, but then no pml file is created for the period when windows is resuming from hybrid sleep.

Would you be kind enough to look at the xml file created by xbootmgr and let me know if you see anything that might help with my problem. Thanks in advance.

Attached File(s)

•  summary_hibernate.xml (286.68K)
  Number of downloads: 3

[MagicAndre1981]

levelblade, on 03 April 2012 - 04:59 PM, said:

Per your instructions, I have run a hibernation trace using xbootmgr and similar to some other users, it shows acceptable hiberwrite time (approx. 21 secs) but very, very long hiberread time (more than 276 secs or over 4 minutes).

yes, reading the hibernation file is horrible slow.

levelblade, on 03 April 2012 - 04:59 PM, said:

From reading prior posts in this thread, I am guessing that the problem may lie with my hard drive. It is a Hitachi HDS721075KLA330 and its specs (7200rpm and 32mb cache) suggest that it should have plenty of speed such that the hiberread time should not be so slow.

do you run the HDD in AHCI or IDE mode? Also try to connect the HDD to a different SATA port. Also scan for SMART Errors.

[levelblade]

MagicAndre1981, on 04 April 2012 - 09:01 AM, said:

do you run the HDD in AHCI or IDE mode? Also try to connect the HDD to a different SATA port. Also scan for SMART Errors.

The HDD runs in IDE mode; AHCI does not appear as an option in the BIOS on my Dell XPS435MT. Checked the drive with HDD Tune and it reports that all is OK. Moved the connector from my HDD to a different SATA port on the motherboard, no improvement in reading the hibernation file. Any other ideas?

[MagicAndre1981]

have you installed the latest Intel chipset drivers? Does it also happen if you run Windows without any 4rd party tools? (http://support.micro...om/kb/929135/en)

[twinkies]

hi guys i have reformatted my computer recently and i found my boot times is slow compared before reformatting.
btw im using an intel ssd g2.

i created a trace using the tutorial and i think the culprit is fltmgr.sys but i have no idea how to solve it. Any help would be appreciated.

thanks in advance

files of the trace :
http://dl.dropbox.co...E_IO_INIT_1.etl

file of the summary:
http://dl.dropbox.co...105/summary.xml

[MagicAndre1981]

looks like you have a disk issue, because chkdsk is running:

\??\C:\Windows\system32\autochk.exe *

This takes 32s and that's why your Windows boots 45s. Check your ST3500410AS, Hitachi HDS5C3030ALA630, Hitachi HDP725025GLA380 or your INTEL SSD for SMART issues. If chkdsk doesn't detect any errors, delete the autock.exe entry in the registry.

[levelblade]

MagicAndre1981, on 05 April 2012 - 10:43 AM, said:

have you installed the latest Intel chipset drivers? Does it also happen if you run Windows without any 4rd party tools? (http://support.micro...om/kb/929135/en)

Used the driver checkup tool on the Intel site, and it reported that all of my Intel drivers are up to date. Went through the whole process of disabling non-MS services and startups via msconfig; no improvement when all third party services are disabled.

I am out of town and will not be using the problem system for the next couple of months. When I get back, I will likely invest in a new, higher performance HDD and see if that resolves the problem. If that doesn't improve the hiberread time, I plan to ditch the Dell system and build a new one (I've already had to replace the inadequate PSU and deficient wireless adapter, so I already have two components for the new system). I will let you know if there is any improvement when I get around to swapping out the HDD.

While I haven't gotten any improvement in the system yet, the process has been informative. Thanks for your help.